As generative artificial intelligence (GenAI) reshapes educational priorities, governance and pedagogical norms, global citizenship education (GCE) in higher education (HE) faces intensifying tensions between ethical imperatives and algorithmic rationalities. This article offers a human-centred, critical-decolonial reframing of GCE in the era of artificial intelligence (AI). It argues that the dominant approaches to GCE risk reinforcing neoliberal, technocratic and universalist assumptions. Drawing on the work of Vanessa Andreotti, Sharon Stein, Emiliano Bosio and Paulo Freire, the article advocates for an alternative grounded in relational ethics, epistemic plurality and institutional reflexivity. Employing critical interpretive synthesis and guided by the performative tool of social cartography, the authors developed a ten-dimension framework for institutional self-reflection. Rather than framing GCE as a fixed set of global competencies, this framework reimagines it as a dynamic pedagogical project grounded in care, shared responsibility and ethical dialogue. It responds to the structural inequities amplified by AI integration, foregrounds the risks of AI-mediated depersonalisation and invites HE institutions to implement GCE as a dynamic space for justice-oriented learning and collective flourishing.